Grok 4.1 Fast: benchmark results
xAI's speed-oriented Grok 4.1 variant for agentic tool-calling workloads with a 2M-token context. Provider: xAI. Released 2025-11-19. Access: API.
Aggregate Rating 1691 ± 4, rank #217 of 1656 rated models, from 1251 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AgencyBench V2 - MCP | 68.6 | Average Task Score (%) | 100 |
| AgentEHR (ReAct) | 25.01 | F1 (%; mean of six clinical prediction tasks on the MIMIC-IV | 100 |
| AgentEHR (ReAct) - Microbiology | 20.32 | F1 (%; predicted microbiology label set against the next rec | 100 |
| AgentEHR (ReAct) - Prescriptions | 16.88 | F1 (%; predicted prescriptions label set against the next re | 100 |
| AgentEHR (ReAct) - Procedures | 37.62 | F1 (%; predicted procedures label set against the next recor | 100 |
| BALSAM - Program Execution | 95.79 | Overall score (0-100, LLM-judged generation and multiple cho | 100 |
| Eye-Q - Partial Character Reveal - English | 60.27 | Exact-Match Accuracy (%; 25% of the answer's characters reve | 100 |
| Hack-Verifiable TextArena | 28.5 | Avg HR (self-reported) | 100 |
| InteractiveBench Math | 76.9 | Accuracy (%) | 100 |
| Ko-AgentBench - L5 Error Handling & Robustness | 34.75 | Adaptive Routing Score (%) | 100 |
| LingxiDiagBench (Dynamic, APA-Guided + MRD-RAG) - 12-Class Accuracy | 28.5 | Accuracy (%, 0-100) on twelve-category ICD-10 diagnosis (F20 | 100 |
| LingxiDiagBench (Dynamic, APA-Guided + MRD-RAG) - 4-Class Accuracy | 43 | Accuracy (%, 0-100) on four-way classification (pure depress | 100 |
Interactive version: theaggregate.ai/model?slug=grok-4-1-fast · How It Works · Data refreshed daily, snapshot 2026-10-20.